Interactions between in vivo neuronal‐glial markers, side of hippocampal sclerosis, and pharmacoresponse in temporal lobe epilepsy

Abstract

Objective

To evaluate the interactions of metabolic neuronal‐glial changes with the presence and hemispheric‐side of hippocampal sclerosis (HS) and its potential role in predicting pharmacoresistance in temporal lobe epilepsy (TLE).

Methods

We included structural magnetic resonance imaging (MRI) and proton magnetic resonance spectroscopy (1H‐MRS) metabolic data for 91 patients with unilateral TLE and 50 healthy controls. We measured the values of total N‐acetyl aspartate/total creatine (tNAA/tCr), glutamate/tCr (Glu/tCr), and myo‐inositol/tCr (mIns/tCr). To assess the influence of the pharmacoresponse and hemispheric‐side of HS on metabolic data, the relationship between clinical and MRI data, and the predictive value of NAA/Cr, we used analysis of variance/covariance and built a logistic regression model. We used bootstrap simulations to evaluate reproducibility.

Results

Bilateral tNAA/tCr reduction was associated with pharmacoresistance and with left HS, a decrease of Glu/tCr ipsilateral to the seizure focus was associated with pharmacoresistance, and ipsilateral mIns/tCr increase was related to pharmacoresistance and the presence of left HS. The logistic regression model containing clinical and 1H‐MRS data discriminated pharmacoresistance (area under the curve [AUC] = 0.78). However, the reduction of tNAA/tCr was the main predictor, with the odds 2.48 greater for pharmacoresistance.

Significance

Our study revealed a spectrum of neuronal‐glial changes in TLE, which was associated with pharmacoresistance, being more severe in left‐sided HS and less severe in MRI‐negative TLE. These noninvasive, in vivo biomarkers provide valuable additional information about the interhemispheric differences in metabolic dysfunction, seizure burden, and HS, and may help to predict pharmacoresistance.

0